rU^^cJ (i The Auk 110(4):825-831, 1993 MITOCHONDRIAL-DNA VARIATION AND EVOLUTIONARY RELATIONSHIPS IN THE AMAKIHI COMPLEX CHERYL L. TARR'-^ AND ROBERT C. FLEISCHER' Department of Biology, University of North Dakota, Grand Forks, North Dakota 58202, USA ABSTRACT.?An analysis of restriction-site variation in mitochondrial DNA was conducted to examine relationships among five taxa in one group of honeycreepers?the amakihi com- plex (genus Hemignathus). We analyzed 35 ingroup and 3 outgroup samples. Tree topologies, based on both distance and parsimony methods, grouped taxa into two distinct lineages: the virens-wilsoni lineage; and the chloris-stejnegeri-parvus group. Inter-island sequence divergence (average d,, = 0.0368) is considerably higher than intra-island variation (mean d, = 0.0035), and is higher than average for avian species. Variability (measured as both nucleotide diversity and maximum divergence between haplotypes) differs among island populations. Molecular evolutionary rates were calibrated on the basis of maximum island age estimates; sequence divergence in this lineage is approximately 2% per million years. The relationships within the chloris-stejnegeri-parvus clade generally are consistent with the previously proposed model of double invasion. Genetic distances and the pattern of relationships among amakihi taxa indicate that species status for H. v. chloris and H. v. stejnegeri may be warranted. Received 25 July 1992, accepted 18 December 1992. THE HAWAIIAN ARCHIPELAGO is widely herald- ed as a "natural laboratory" for evolutionary studies (Carlquist 1980, Simon 1987). The iso- lation and ecological diversity of the Hawaiian Islands have served as the backdrop for nu- merous adaptive radiations. As an example of adaptive radiation of an insular avifauna, the Hawaiian honeycreepers (Drepanidinae) are unsurpassed. This group is renowned for the morphological, ecological, and behavioral di- versity that exists among the 23 species de- scribed historically (Amadon 1950, Bock 1970, Freed et al. 1987). The recent description of an additional 14 species (James and Olson 1991) has shown that this diversity was once even more extensive. The drepanidine radiation has been attrib- uted to recurrent double colonization (Amadon 1950, Bock 1970). This process involves colo- nization of a new island, then differentiation and development of reproductive isolation dur- ing a period of allopatry. This is followed by competition and character displacement when secondary contact occurs. Double colonization has been proposed as a general mechanism for ' Present address: Molecular Genetics Program, Na- tional Zoological Park, Smithsonian Institution, Washington, D.C. 20008, USA. ^ Department of Biology and Institute of Molecular Evolutionary Genehcs, Pennsylvania State Universi- ty, University Park, Pennsylvania 16802, USA. speciation of island birds (Lack 1947, Grant 1981). Patterns of colonization consistent with this hypothesis have not been documented for the honeycreepers, in part because phyloge- netic relationships of this group have not been fully resolved. In addition to a corroborated phylogenetic hypothesis, an accurate time frame for the diversification of this group needs to be established. In order to address these issues, we have con- ducted an analysis of restriction-site variation in one group of closely related honeycreepers of the genus Hemignathus, known as the amakihi complex. Speciation via double invasion is thought to have occurred tw^ice within the ama- kihi complex (Amadon 1950, Bock 1970). The complex includes the following taxa: Hemigna- thus sagittirostris, once found locally on Hawaii, but now extinct; H. parvus, a species endemic to Kauai; and H. virens, a polytypic species found throughout the main islands. Four subspecies of H. virens have been described: H. v. virens on Hawaii; H. v. wilsoni on Maui, Molokai, and (for- merly) Lanai; H. v. chloris on Oahu; and H. v. stejnegeri on Kauai. Two species of amakihi are found on both Hawaii and Kauai, and it has been proposed that both species pairs have aris- en via double colonization from a neighboring island (Amadon 1950, Bock 1970). The suitability of mitochondrial DNA (mtDNA) for addressing questions of taxonomy and evolutionary relationships among closely 825 826 TARR AND FLEISCHER [Auk, Vol. 110 Kauai t Niitiau ^ r^ 1 . V. steinegari ?1 pan/us OahuVV* "?agi 4 6 H. V. citions i ??^f^-J H. V. nilsoni KahoolaweOy N M g j ^^ H. V. wrens 9^^^ -VHawaii 0 100 "~V/r Fig. 1. Collecting localities in the Hawaiian Is- lands: (1) Pihea trail (n = 7, H. v. stejnegeri; ? = 5, H. parvus); (2) Keaiwa Heiau (? = 2); (3) Makiki (? = 2); (4) West Maui (n = 2); (5) Polipoli (? = 3); (6) Puu Alaea {n = 1); (7) Parker Ranch (n = 2); (8) Puu Laau (n = 2); (9) Puu O Kauha {n = 3); (10) Hualalai (n = 1); (11) Manuka (w = 3); (12) Kulani (? = 1); (13) Puu Kanakaleonui (n = 1). related organisms has been extensively re- viewed (e.g. Avise et al. 1987, Wilson et al. 1985, Shields and Helm-Bychowski 1988, Harrison 1989). Here we report on an analysis of mtDNA variation in five taxa in the amakihi complex. Our goals are to: (1) estimate a phylogeny for the amakihi complex and infer from this the pattern and timing of island colonization; and (2) provide a calibration of nucleotide substi- tution rates based on estimates of island age. In addition, we reinterpret the classification of H. virens in light of the phylogeny and genetic diversity measures. METHODS We collected 35 ingroup samples from four of the main Hawaiian Islands. Specific localities and sample sizes are shown in Figure 1. Three individuals of the Laysan Finch {Telespiza cantans) were included as an outgroup species; this species was chosen because it represents a lineage that arose earlier than the ama- kihis in the honeycreeper radiation (Tarr and Fleisch- er in press). Liver, heart, and breast muscle were re- moved from specimens, stored in MSB buffer (Lansman et al. 1981), and frozen in liquid nitrogen. Samples were later stored at ? 90?C for up to two years. A modification of the sucrose step gradient protocol (Spolsky and Uzzell 1984) was developed and used to prepare mtDNA (Tarr 1991). All mtDNA samples were digested with 15 six-base-recognizing restric- tion endonucleases (BamHl, Banlll, Bc/I, Bgll, BglU, BsfBl, EcoRl, EcoT221, Hindlll, Kpnl, Psil, Sad, Sail, Xba?, Xhol). For samples without any contaminating nuclear DNA, about 20 ng per sample was digested and end-labelled with '^P and electrophoresed on 1.0 or 1.2% agarose gels at 70 to 90 V for 12 to 18 h. A molecular size standard (lambda DNA cut with Hmdlll) was included for determination of fragment sizes. Gels were dried under vacuum and fragments were visualized by autoradiography. Some tissue samples (n = 11) had been thawed and refrozen prior to mtDNA isolation and yielded mtDNA that was contaminated with nuclear DNA. These sam- ples were digested and electrophoresed as above. The DNA in the gel was depurinated (0.25 N HCl), fol- lowed by two denaturing washes (1.5 M NaCl:0.5 N NaOH), and the DNA was transferred in 20 x SSC to an MSI Magna NT nylon membrane with a vacuum blotter. The membranes were prehybridized in 4x SSC, 5% sodium pyrophosphate, 1 x Denhardt's so- lution, and 0.5% SDS for 6 h at 65?C. Approximately 100 ng of House Finch (Carpodacus mexicanus) mtDNA was labelled with '^PdATP by the random-primed hexamer method to a specific activity of 0.7-2.4 x 10' cpm/Mg. Hybridizations were carried out in a shaking water bath at ?S'C for 18 to 24 h. Filters were washed twice (2x SSC, 0.2% SDS, 0.1 x Denhardt's) at room temperature for 15 min, followed by two 20-min washes (2x SSC, 0.1% SDS) at 65?C. One to three individuals were analyzed by both of the above meth- ods for 11 of the 15 enzymes. Following the approach of Kessler and Avise (1984), a matrix was constructed by scoring the presence or absence of fragments for each individual. Compari- sons of fragment sizes were used to infer the mini- mum number of restriction-site changes that could account for the changes in fragment profiles, and a presence/absence matrix of sites was constructed. The proportion of shared DNA fragments (F), nu- cleotide diversity within populations (d,), and the average number of nucleotide substitutions per site between haplotypes from two populations (d,^) were calculated by the computer program RESTSITE (Mil- ler 1990, Nei and Miller 1990), which uses formulae 5.53, 5.54, and 5.55 from Nei (1987). Similar measures of diversity were calculated for sites (Nei 1987: for- mulae 5.38 and 5.42). A range of substitution rates was estimated by dividing d^^ (uncorrected and cor- rected values) by the age of the oldest volcanic series for a given island (this assumes an island was colo- nized soon after emergence and, thus, provides a min- imum rate). The UPGMA option in the RESTSITE program (Mil- ler 1990) was used to cluster the taxa based on distance estimates. The neighbor-joining algorithm (Saitou and Nei 1987) in the analysis package MEGA (Kumar et al. 1993) also was used to construct a tree from the distance matrix. A parsimony analysis was performed by the program PAUP (Swofford 1985) with T. cantans as an outgroup to root the tree. A bootstrap was per- formed on the site matrix (branch-and-bound algo- rithm with 250 replications). October 1993] Variation in Amakihi intDNA 827 TABLE 1. Nucleotide diversity in four subspecies of H. virens. Taxon SE Maximum d virens 0.00149 ? 0.00077 0.00460 wilsoni 0.00468 ? 0.00176 0.01255 Moris 0.00197 ? 0.00135 0.00425 stejnegeri 0.00520 ? 0.00299 0.00750 RESULTS The average size of the mtDNA molecule in Hemignathus is approximately 16,712 ? SD of 363 bp. No size variation was detected in this study. The 15 restriction endonucleases pro- duced 115 fragments, 15 of which were unique to T. cantans. We identified 18 haplotypes among the 38 individuals; 16 of these were in H. virens (there was only one haplotype each in H. parvus and T. cantans). We estimated 73 sites from the fragment profiles. All enzymes showed restric- tion-site variation among the taxa, and all ex- cept SflcII revealed at least one polymorphic restriction site within H. virens. The analyses based on sites and fragments did not differ greatly (Tarr 1991), and only the results based on the site data are presented in detail here. Estimates of intra-island diversity for H. vi- rens indicate that variability differs among is- land populations (Table 1). The greatest degree of differentiation was between H. v. wilsoni hap- lotypes. Two haplotypes from one location (Po- lipoli. Maul) differed by six restriction sites (d = 0.0122). The maximum sequence divergence between H. v. stejnegeri haplotypes also was high, with five restriction-site differences. Haplo- types of H. V. virens and H. v. chloris differed by no more than two restriction sites, and the max- imum sequence divergence was less than 0.5% (Table 1). The average sequence divergence among is- land populations of H. virens, based on analysis li '?3 ^\ H. V. wilsoni H. V. stejnegeri H. parvus T. canlans ?r- 5.0 ?r- 1.0 Percent sequence divergence Fig. 2. UPGMA phenogram produced by analysis of restriction-site distances. of sites, was 0.0370 ? 0.012. Hemignathus v. virens and H. v. wilsoni were the most closely related taxa, and the highest divergence was between H. V. chloris and H. v. wilsoni (Table 2). The UPGMA analysis shows island popula- tions as distinct clusters (Fig. 2). Two major lin- eages are evident: the virens-wilsoni group; and the parvus-stejnegeri-chloris group. Within the latter lineage, H. v. parvus and H. v. stejnegeri form a cluster. The bootstrap consensus tree of 76 steps is shown in Figure 3. The neighbor-joining al- gorithm produced a topology identical to the parsimony tree (Fig. 4). Most features of the topology are concordant with the phenogram in Figure 2. For example, all three analyses par- tition the amakihis into two distinct lineages; and each island population is a monophyletic group. However, the position of H. parvus dif- fers in the analyses: H. v. chloris and H. parvus are sister taxa in the cladogram and neighbor- joining tree; and H. v. stejnegeri is the sister tax- on to H. parvus in the UPGMA phenogram. Also, TABLE 2. Average sequence divergence (d,,) between taxa (above diagonal) with jackknifed standard errors (below diagonal). Taxon Taxon 1 2 3 4 5 6 1 H. V. virens ? 0.0205 0.0475 0.0373 0.0448 0.0473 2 H. V. wusoni 0.0055 ? 0.0551 0.0440 0.0476 0.0609 3 H. V. chloris 0.0088 0.0094 ? 0.0374 0.0416 0.0851 4 H. V. stejnegeri 0.0092 0.0074 0.0107 ? 0.0357 0.0626 5 H. parvus 0.0077 0.0090 0.0139 0.0121 ? 0.0573 6 T. cantans 0.0113 0.0135 0.0192 0.0137 0.0132 ? 828 TARR AND FLHSCHER [Auk, Vol. 110 62 26 95 58 59 63 24 1 1 47 100 65 1 1 96 36 ? 82 2 1 3 4 5 4 1 2 3 2 3 1 1 H.parws 1 H. V. stejnggeri T. cantans Fig. 3. Bootstrap consensus tree produced by par- simony analysis of presence/absence matrix of sites. Numbers along branches denote percentages of trees in which a particular node appears. relationships among haplotypes within islands are not consistently resolved. DISCUSSION mtDNA diversity in amakihi populations.?Di- versity mccisures indicate average to high levels of intra-island variation in populations of H. virens. The nucleotide diversity in the Hawaii and Oahu populations, although low compared to Kauai and Maui, are similar to values re- ported for other avian taxa (Avise and Zink 1988, Seutin et al. 1993). In contrast, nucleotide di- versity is high in the Maui and Kauai popula- tions, and in some cases the pairwise divergence between individuals is much higher than val- ues normally reported for birds collected from the same site, and even exceed the divergence between some congeneric species of birds (e.g. Avise and Zink 1988:tables 4 and 5). The percent sequence divergence between two individuals from Polipoli (1.22%) is as high as the value reported between two subspecies of Platycercus elegans; Ovenden et al. (1987) reported a max- imum nucleotide divergence of 1.17% between individuals of P. e. nigrescens from the north- eastern coast of Australia and P.e. adelaidae from the south-central coast of Australia. One other example of co-occurring divergent lineages within a population of birds has been reported; a study of mtDNA polymorphism in popula- tions of Parus caeruleus in France (Taberlet et al. 1992) revealed two maternal lineages that dif- fered by 1.23%. The phylogeographic pattern of ? 3 ?4 .5 '-2 ^ ^ H. V. ste/negari -H.parvus -2 \ 3 I - I - r. cantons Fig. 4. Distance-parsimony tree constructed by neighbor-joining algorithm. P. caeruleus was attributed to either recent ad- mixture of two previously isolated populations, or introgression through hybridization with P. cyanus. Since the mitochondrial genome is haploid and maternally inherited, its effective popula- tion size is about four times smaller than that for nuclear genes (Nei 1987). Thus, mtDNA variation can be lost more rapidly through pop- ulation reduction than nuclear-DNA variation (Wilson et al. 1985). The model of Avise et al. (1984) predicts that, unless the number of gen- erations is small or N^ is very large, the prob- ability of maintaining more than one lineage is very low. However, the probability of survival of two or more mtDNA lineages is increased if a population is subdivided (Avise et al. 1984). Although mtDNA diversity will decay within each subpopulation, at least one mtDNA lin- eage per deme will be maintained. Occasional migration between demes could introduce moderately divergent mtDNA lineages into a population. Levels of variability indicate that long-term effective population sizes have varied among amakihi populations; the factors causing such variation are unknown. However, population subdivision may play a role in maintaining di- versity in at least one deme. The islands of Maui, Molokai, and Lanai were joined to form the Maui Nui complex during the Wisconsin glacial period, and may have been contiguous during previous glacial periods (Stearns 1966, Macdon- ald et al. 1983). Secondary contact between pre- viously isolated demes could occur during low- er stands of the sea and, thereby, enhance genetic diversity. The monophyly of island populations October 1993] Variation in Amakihi intDNA 829 and the presence of intermediate haplotypes makes subspecies hybridization an unlikely source of divergent haplotypes in H. v. wilsoni. Rate of mtDNA evolution and island coloniza- tion. ?Some authors have suggested the average rate of substitution over the mitochondrial ge- nome may be decelerated in birds compared to other vertebrates (Kessler and Avise 1985, Ovenden et al. 1987; but see Shields and Wilson 1987). Because the ages of the Hawaiian Islands impose maximum ages on the populations, the minimum substitution rate of the mtDNA ge- nome can be estimated. We use only the two youngest islands because the time of diver- gence of the amakihis from other drepanidines is not known; the time may be well accom- modated within the ages of Kauai and Oahu. Assuming a substitution rate of 2% per mil- lion years (Shields and Wilson 1987), the Ha- waii subspecies (H. v. virens) and the Maui sub- species (H. V. wilsoni) diverged approximately 1,000,000 years ago. If the most recent age es- timate for Hawaii (400,000 years; McDougall and Swanson 1972) is accurate, the divergence event would predate the formation of Hawaii by about 600,000 years. The branch leading to the Maui and Hawaii populations diverged from the Kau- ai-Oahu lineage approximately 2.2 mybp. This would predate the formation of Maui (1.6 mybp; Naughton et al. 1980) by about 600,000 years. Assuming Hawaii and Maui were colonized in the order of formation, we used the ages of the islands to calibrate a substitution rate of 2.7 and 5.0% sequence divergence per million years for the Maui and Hawaii lineages, respectively. However, because Maui and Molokai have been joined in the past, the age of Molokai (1.84 my; Naughton et al. 1980) may be more appropriate for calibrating a substitution rate for H. v. wil- soni; this yields a rate of substitution of 2.4% per million years. The above calibration does not account for polymorphism at the time of population split- ting. Haplotypes could diverge in the ancestral population prior to inter-island dispersal, thereby leading to an overestimate of the ge- netic distance between the two populations. Such an overestimate could be extreme in the case of H. v. wilsoni and H. v. virens, because divergent lineages presently exist on Maui, and may also have existed in the past. We maintain that one population was derived from the other (on basis of close genetic relationship between H. V. virens and H. v. wilsoni); the direction of colonization was most likely from Maui to Ha- waii (if Maui were colonized from Hawaii, then a rate of 10% divergence per million years would be required to account for the distance between the virens-wilsoni and chloris-stejnegeri lineages). To provide a maximum correction for estimat- ing the net nucleotide substitutions per site {dy, Nei 1987), we subtract the maximum diver- gence between haplotypes on Maui (0.012) from d? (0.020), which yields a d^ of 0.008, and a substitution rate of approximately 2% per mil- lion years. As the potassium-argon ages for Ha- waii are equivocal (e.g. McDougall 1969, Dal- rymple 1971), the calibration for H. v. virens is not as reliable as that for H. v. wilsoni. We have made conservative assumptions to provide a minimum rate in our calibrations, and we con- clude that the rate of molecular evolution in this group of birds is not decelerated relative to other vertebrates. We also note that some rate heterogeneity exists among lineages of H. virens (Fig. 4), and the substitution rate may be accel- erated in the populations on the two older is- lands. The phylogeny does not provide further in- sights into patterns of colonization (beyond those inferred from comparisons of genetic dis- tance and island age). We cannot be certain whether the Kauai or Oahu population is an- cestral, and the position of H. parvus has proven difficult to resolve (see also Johnson et al. 1989). The topology derived from the parsimony and neighbor-joining analyses is consistent with secondary contact of taxa on Kauai, as suggested by Amadon (1950) and Bock (1970); a neighbor- joining analysis, which includes other drepan- idine taxa, supports the double-invasion hy- pothesis as well (Tarr and Fleischer in press). However, the UPGMA tree does not support the double-invasion hypothesis, but instead is consistent with a sympatric origin for H. parvus and H. v. stejnegeri; a parsimony analysis (Tarr and Fleischer in press) is inconsistent with in- clusion of H. parvus in the amakihi complex. Future studies which include H. v. sagtttirostris and other drepanidine taxa will provide a more rigorous test of the double invasion hypothesis. Taxonomic considerations.?The taxonomic sta- tus of H. V. stejnegeri has been debated, primarily because of its morphological divergence from other populations of H. virens. The mtDNA dif- ferentiation of the Kauai Amakihi reported here is consistent with previous investigations. Mor- phological (Amadon 1950, Bock 1970, Pratt 1979), 830 TARR AND FLEISCHER [Auk, Vol. 110 osteological (James and Olson 1991), ecological and behavioral (Pratt 1979) differentiation, as well as nuclear differentiation (Johnson et al. 1989), all indicate that the divergence of H. v. stejnegeri is sufficient to consider it a separate species. The nomenclature for the Kauai Ama- kihi remains problematic, as the epithet H. stejnegeri is preoccupied (Olson and James 1988); Pratt (1989) has suggested the name H. kauaien- sis. However, James and Olson (1991) have re- tained the amakihis in the genus Loxops, with the Kauai Amakihi designated as Loxops stejne- geri. A molecular-phylogenetic analysis of the Hawaiian honeycreepers will provide an in- dependent assessment of the relationship of the amakihis to other drepanidines (Fleischer, Mc- Intosh, and Tarr pers. comm.). Surprisingly, our data also suggest the diver- gence of H. V. chloris may be sufficient to con- sider it a separate species. The sequence diver- gence between the Hawaii-Maui clade and the Kauai-Oahu clade (4.4%) is similar to the aver- age distance among taxa in five avian genera (Avise and Zink 1988). The divergence between the Kauai and Oahu amakihi (3.4%) also is sim- ilar to levels of interspecific differentiation in other avian taxa. Some authors caution against delineating species limits on the basis of a gene tree (e.g. Quinn et al. 1991). Examination of nuclear genes can reveal different patterns of divergence, and there is no absolute value of divergence above which two taxa should be considered separate species. Hemignathus v. chloris shows a slight to moderate divergence in nuclear genes (Johnson et al. 1989) and morphological differentiation appears slight. However, if the phylogeny pre- sented here accurately reflects the pattern of descent among the amakihis, then H. virens as it is currently defined is polyphyletic. The case for separating H. v. chloris is not as strong as that for H. v. stejnegeri, but these data are sug- gestive, and further analysis seems warranted. ACKNOWLEDGMENTS We thank C. Aguon, R. Clapp, S. Conant, F. Duvall, A. Engelis, L. Freed, H. James, C. Mclntosh, L. Miller, T. Pratt, E. Rave, G. Seutin, and R. Williams for as- sistance in the field, and K. Gibson, L. Miller and J. Werner for assistance in the laboratory. R. Crawford, J. LaDuke, S. Olson, G. Shields, and an anonymous reviewer provided constructive comments on earlier drafts of the manuscript. We thank the following for providing logistic support and/or permissions: S. Co- nant; J. Jacobi and T. Pratt (U.S. Fish and Wildlife Service); C. Terry (Department of Forestry and Wild- life, Hawaii Department of Land and Natural Re- sources); C. Stone (Hawaii Volcano National Park); M. Morin; A. Holt and E. Misaki (The Nature Con- servancy); Maul Land and Pineapple Co.; and the Par- ker Ranch. The National Geographic Society, U.S. Fish and Wildlife Service (Contract 14-16-0005-90-027), Sigma Xi, Hawaii Audubon Society, Frank M. Chap- man Memorial Fund, and University of North Dakota provided funding for this research. LITERATURE CTTED AMADON, D. 1950. The Hawaiian honeycreepers (Aves, Drepaniidae). Bull. Am. Mus. Nat. Hist. 95:155-262. AVISE, J. C, J. E. NEIGEL, AND J. ARNOLD. 1984. De- mographic influences on mltochondrial DNA lineage survivorship in animal populations. J. Mol. Evol. 20:99-105. AVISE, J. C, J. ARNOLD, R. M. BALL, E. BERMINGHAM, T. LAMB, J. E. NEIGEL, C. A. REEB, AND N. C. SAUNDERS. 1987. Intraspecific phylogeography: The mitochondrial bridge between population genetics and systematics. Annu. Rev. Ecol. Syst. 18:489-522. AVISE, J. C, AND R. M. ZINK. 1988. Molecular genetic divergence between avian sibling species: King and Clapper rails. Long-billed and Short-billed dowitchers. Boat-tailed and Great-tailed grackles, and Tufted and Black-crested titmice. Auk 105: 516-528. BOCK, W. J. 1970. Microevolutionary sequences as a fundamental concept in macroevolutionary mod- els. Evolution 24:704-722. CARLQUIST, S. 1980. Hawaii: A natural history. Nat- ural History Press, Garden City, New York. DALRYMPLE, G. B. 1971. Potassium-argon ages from the Pololu volcanic series, Kohola volcano, Ha- waii. Geol. Soc. Am. Bull. 82:1997-1999. FREED, L. A., S. CONANT, AND R. C. FLEISCHER. 1987. Evolutionary ecology and radiation of Hawaiian passerine birds. Trends Ecol. & Evol. 2:196-203. GRANT, P. R. 1981. Ecology and evolution of Dar- win's finches. Princeton Univ. Press, Princeton, New Jersey. HARRISON, R. G. 1989. Animal mitochondrial DNA as a genetic marker in population and evolution- ary biology. Trends Ecol. & Evol. 4:6-11. JAMES, H. F., AND S. L. OLSON. 1991. Descriptions of thirty-two new species of birds from the Hawai- ian Islands: Part II. Passeriformes. Ornithol. Monogr. 46. JOHNSON, N. K., J. A. MARTEN, AND C. J. RALPH. 1989. Genetic evidence for the origin and relationships of Hawaiian honeycreepers (Aves: Frlngillidae). Condor 91:379-396. KESSLER, L. G., AND J. C. AVISE. 1984. Systematic re- October 1993] Variation in Amakihi mtDNA 831 lationships among waterfowl (Anatidae) inferred from restriction endonuclease analysis of mito- chondrial DNA. Syst. Zool. 33:370-380. KESSLER, L. G., AND J. C. AVISE. 1985. A comparative description of mitochondrial DNA differentia- tion in av?an and other vertebrate genera. Mol. Biol. Evol. 2:109-125. KUMAR, S., K. TAMURA, AND M. NEI. 1993. MEGA: Molecular evolutionary genetics analysis pack- age, v. 1.00. Pennsylvania State Univ., State Col- lege, Pennsylvania. LACK, D. 1947. Darwin's finches. Cambridge Univ. Press, New York. LANSMAN, R. A., R. O. SHADE, J. F. SHAPIRA, AND J. C. AVISE. 1981. The use of restriction endonucleas- es to measure mitochondrial DNA sequence re- latedness in natural populations. 111. Techniques and potential applications. J. Mol. Evol. 17:214- 226. MACDONALD, G. A., A. T. ABBOTT, AND F. L. PETERSON. 1983. Volcanoes in the sea. Univ. Hawaii Press, Honolulu. McDoUGALL, I. 1969. Potassium-argon ages on lavas of Kohala volcano, Hawaii. Geol. Soc. Am. Bull. 80:2597-2600. McDouGALL, I., AND D. A. SwANSON. 1972. Potas- sium-argon ages of lavas from the Hawi and Po- lolu volcanic series, Kohala Volcano, Hawaii. Geol. Soc. Am. Bull. 83:3731-3738. MILLER, J. C. 1990. Programs for analyzing restric- tion site or fragment data. Univ. Wisconsin, Mad- ison. NAUGHTON, J. J., G. A. MACDONALD, AND V. A. GREEN- BERG. 1980. Some additional potassium-argon ages of Hawaiian rocks: The Maui volcanic com- plex of Molokai, Maui, Lanai and Kahoolawe. J. Volcanol. Geotherm. Res. 7:339-355. NEI, M. 1987. Molecular evolutionary genetics. Co- lumbia Univ. Press, New York. NEI, M., AND J. C. MILLER. 1990. A simple method for estimating average number of nucleotide sub- stitutions within and between populations from restriction data. Genetics 125:873-879. OLSON, S. L., AND H. F. JAMES. 1988. Nomenclature of the Kauai Amakihi and Kauai Akialoa (Dre- panidini). 'Elepaio 48:13-14. OVENDEN, J. R., A. G. MACKINLAY, AND R. H. CROZIER. 1987. Systematics and mitochondrial genome evolution of Australian rosellas (Aves: Platycer- cidae). Mol. Biol. Evol. 4:526-543. PRATT, H. D. 1979. A systematic analysis of the en- demic avifauna of the Hawaiian Islands. Ph.D. dissertation, Louisiana State Univ., Baton Rouge. PRATT, H. D. 1989. A new name for the Kauai Ama- kihi (Drepanidinae: Hemignalhus). 'Elepaio 49: 13-14. QUINN, T. W., G. F. SHIELDS, AND A. C. WILSON. 1991. Affinities of the Hawaiian Goose based on two types of mitochondrial DNA data. Auk 108:585- 593. SAITOU, N., AND M. NEI. 1987. The neighbor-joining method: A new method for reconstructing phy- logenetic trees. Mol. Biol. Evol. 4:406-425. SEUTIN, G., J. BRAWN, R. E. RICKLEFS, AND E. BERMING- HAM. 1993. Genetic divergence among popu- lations of a tropical passerine, the Streaked Sal- tator {Saltator albicollis). Auk 110:117-126. SHIELDS, G. F., AND K. M. HELM-BYCHOWSKI. 1988. Mitochondrial DNA of birds. Curr. Ornithol. 5:273-295. SHIELDS, G. F., AND A. C. WILSON. 1987. Calibration of mitochondrial DNA evolution in geese. J. Mol. Evol. 24:212-217. SIMON, C. 1987. Hawaiian evolutionary biology; An introduction. Trends Ecol. & Evol. 2:175-178. SPOLSKY C, AND T. UZZELL. 1984. Natural interspe- cies transfer of mitochondrial DNA in amphib- ians. Proc. Nati. Acad. Sei. USA 81:5802-5805. STEARNS, H.T. 1966. Geology of the State of Hawaii. Pacific Books, Palo Alto, California. SWOFFORD, D. 1985. Phylogeneticanalysis using par- simony. Illinois Natural History Survey, Cham- paign-Urbana. TABERLET, P., A. MEYER, AND J. BOUVET. 1992. Un- usual mitochondrial DNA polymorphism in two local populations of Blue Tit Parus caeruleus. Mol. Ecol. 1:27-36. TARR, C. L. 1991. Mitochondrial DNA variation in the 'amakihi complex. M.S. thesis, Univ. North Dakota, Grand Forks. TARR, C. L., AND R. C. FLEISCHER. Evolutionary rela- tionships of the Hawaiian honeycreepers (Aves: Drepanidinae). In press. In Patterns of speciation and biogeography of Hawaiian biota (W. L. Wag- ner and V. A. Funk, Eds.). Smithsonian Institu- tion Press, Washington, DC. WILSON, A. C, R. L. CANN, S. M. CARR, M. GEORGE, U. B. GYLLENSTEN, K. M. HELM-BYCHOWSKI, R. G. HIGUCHI, S. R. PALUMBI, E. M. PRAGER, R. D. SAGE, AND M. STONEKING. 1985. Mitochondrial DNA and two perspectives on evolutionary genetics. Biol. J, Linn. Soc. 26:375-400.